Android is built with multiple layers of security and privacy protections to help keep you, your devices, and your data safe. Most importantly, we are committed to transparency, so you can see your device safety status and know how your data is being used.
Android uses the best of Google’s AI and machine learning expertise to proactively protect you and help keep you out of harm’s way. We also empower you with tools that help you take control of your privacy.
I/O is a great moment to show how we bring these features and protections all together to help you stay safe from threats like phishing attacks and password theft, while remaining in charge of your personal data.
Safe Browsing: faster more intelligent protection
Android uses Safe Browsing to protect billions of users from web-based threats, like deceptive phishing sites. This happens in the Chrome default browser and also in Android WebView, when you open web content from apps.
Safe Browsing is getting a big upgrade with a new real-time API that helps ensure you’re warned about fast-emerging malicious sites. With the newest version of Safe Browsing, devices will do real-time blocklist checks for low reputation sites. Our internal analysis has found that a significant number of phishing sites only exist for less than ten minutes to try and stay ahead of block-lists. With this real-time detection, we expect we’ll be able to block an additional 25 percent of phishing attempts every month in Chrome and Android1.
Safe Browsing isn’t just getting faster at warning users. We’ve also been building in more intelligence, leveraging Google’s advances in AI. Last year, Chrome browser on Android and desktop started utilizing a new image-based phishing detection machine learning model to visually inspect fake sites that try to pass themselves off as legitimate log-in pages. By leveraging a TensorFlow Lite model, we’re able to find 3x more2 phishing sites compared to previous machine learning models and help warn you before you get tricked into signing in. This year, we’re expanding the coverage of the model to detect hundred
[…]
Content was cut in order to protect the source.Please visit the source for the rest of the article.
Read the original article: